## Using Stata for Standard Pairwise and Network Meta-Analysis

We hope you will find the material presented useful. If you use the Stata routines please kindly cite the articles that describe them:

**For the metamiss2 command**

**For the metamiss2 command**

*Mavridis D, White IR, Higgins JPT, Cipriani A, Salanti G. Allowing for uncertainty due to missing continuous outcome data in pairwise and network meta-analysis. Stat Med 2015; 34: 721–741*

*White IR, Higgins JPT, Wood A. Allowing for uncertainty due to missing data in meta-analysis—Part 1: Two-stage methods. Stat Med 2008; 27:711–727*

**For the network graphs package**

**For the network graphs package**

*Chaimani A, Higgins JP, Mavridis D, Spyridonos P, Salanti G. Graphical Tools for Network Meta-Analysis in STATA. PLoS One. 2013 Oct 3;8(10):e76654.*

*Chaimani A, Salanti G. Visualizing assumptions and results in network meta-analysis: the network graphs package. Stata Journal 2015; 15(4): 905-950. *

### How to obtain the routines

To install the routines you need to do the following:

**Installation via live internet connection**

With Stata running, the latest versions of the routines can be downloaded by typing in the command window:

`.net from http://www.mtm.uoi.gr`

and clicking at the respective link.

For questions and comments please contact Dr. Anna Chaimani This email address is being protected from spambots. You need JavaScript enabled to view it.

### Description of the routines

**1. The metamiss2 command - Accounting for missing outcome data in meta-analysis**

**1. The metamiss2 command - Accounting for missing outcome data in meta-analysis**

### Background

Missing outcome data are a common threat to the validity of randomized trials and their meta-analysis, as they require making untestable assumptions. Researchers typically ignore missing data and analyze complete data only; this approach is equivalent to assuming that missing participants are missing at random (MAR). The use of informative missingness parameters (IMP) that relate the outcome in the missing data with that in the observed data has been previously suggested for handling missing outcome data in meta-analysis of binary outcomes.

In Stata, information about missing data can be incorporated in meta-analyses of binary outcomes using the metamiss command. Recently, the IMP framework was extended into meta-analyses with continuous outcomes. The metamiss2 command performs a two-stage approach: it first estimates the ‘adjusted’ study-specific relative effects and their variances and covariances, and then calls metan or network meta to obtain the summary effects.

**Latest update of the command: 16 August 2018**

To access the the help file type:

`.help metamiss2`

Note that the command requires the latest versions of metan, mvmeta and network and is compatible with Stata versions 12, 13 and 14.

**2. The network graphs package - Visualizing assumptions and results in network meta-analysis**

**2. The network graphs package - Visualizing assumptions and results in network meta-analysis**

### Background

Although network meta-analysis has been established as a useful evidence synthesis tool it has been often criticized for its complexity and for been accessible only to researchers with strong statistical and computational skills. Careful evaluation of its assumptions and understandable, concise presentation of the results are necessary to avoid misinterpretations and inform decision-making.

We consider the implementation of network meta-analysis in Stata (using the network package available from http://www.mrc-bsu.cam.ac.uk/software/stata-software/) as our starting point and we provide a series of Stata routines that can be used to produce useful graphical and numerical tools to enhance understanding of network meta-analysis procedures and findings.

### Material

You can download here slides, handout, datasets and the .do file from the workshop "Graphs to enhance understanding & improve interpretability of the evidence from network meta-Analysis: a hands-on tutorial in Stata", 23rd Cochrane Colloquium, Vienna, October 2015.

Download data from various applications of Network Meta-Analysis here (.zip file)

### Included commands

networkplot | Plot for networks of interventions in terms of nodes and edges |
---|---|

netweight | Contribution of each direct comparison in network meta-analysis estimates |

ifplot | Evaluation of statistical inconsistency in networks of interventions |

netfunnel | Comparison-adjusted funnel plot for a network of interventions |

intervalplot | Confidence & Predictive intervals plot |

netleague | League table for networks of interventions |

sucra | Ranking plots for a single outcome of network meta-analysis using probabilities of assuming up to a specific rank |

mdsrank | Ranking of treatments in networks of interventions using multidimensional scaling |

clusterank | Clustering for treatments of a network of interventions according to their performance on two outcomes |

**Latest update of the commands: 22 June 2018**

To access the the help files type:

`.help network graphs`

All routines can be also used via dialog boxes by typing 'db' and the name of each routine, for example:

`.db networkplot`

Note that the package requires the latest versions of metan, metareg, mvmeta and network and are compatible with Stata versions 13, 14 and 15.